ICES A
DVISORYC
OMMITTEEICES CM 2011/ACOM:33
Report of the Workshop on Implementing the ICES Fmsy Framework (WKFRAME-2)
10-14 January 2011
ICES, Denmark
*agreed management plan means endorsed by ICES and implemented by a competent authority
International Council for the Exploration of the Sea Conseil International pour l’Exploration de la Mer
H. C. Andersens Boulevard 44–46 DK-1553 Copenhagen V
Denmark
Telephone (+45) 33 38 67 00 Telefax (+45) 33 93 42 15 www.ices.dk
Recommended format for purposes of citation:
ICES. 2011. Report of the Workshop on Implementing the ICES Fmsy Framework (WKFRAME-2), 10-14 February 2011, ICES, Denmark. ICES CM 2011/ACOM:33. 110 pp.
For permission to reproduce material from this publication, please apply to the Gen- eral Secretary.
The document is a report of an Expert Group under the auspices of the International Council for the Exploration of the Sea and does not necessarily represent the views of the Council.
© 2011 International Council for the Exploration of the Sea
Contents
Executive summary ... 1
1 Introduction ... 2
2 Evaluation of the implementation of the MSY approach in the 2010 advice. ... 4
2.1 Overview of MSY reference points already defined ... 4
2.2 A note on the implied function of MSYBtrigger ... 5
2.3 Feedback from ICES EG’s experiences in 2010 ... 6
2.4 Summary of areas identified for further development... 14
2.4.1 Sole meta analysis ... 14
2.4.2 Detecting Regime shifts ... 15
2.4.3 The development of HCR’s for generic species types ... 18
3 Further developments of the ICES MSY approach ... 20
3.1 Choice of methodology to derive Fmsy values. ... 20
3.2 The use of calculated Fmsy values in formulating advice ... 21
3.3 The choice of intervals around Fmsy (target F) and their use ... 21
3.4 Improved guidelines for setting Blim ... 23
3.5 Modification of advice from Fmsy target advice due to stock status ... 24
3.5.1 If SSB is estimated to be above Btrigger ... 24
3.5.2 If SSB estimated below Btrigger and above Blim. ... 24
3.5.3 Transitional Ftargets if SSB < Btrigger ... 25
4 Further develop the MSY approach to be applied in situations where no analytical assessment is available. ... 29
4.1 Updated guidelines for providing advice where there is no analytical assessment. ... 29
4.1.1 Stocks managed by a target escapement strategy ... 29
4.1.2 Other stocks where there is no accepted analytical assessment ... 29
4.1.3 Limitations to the current approach ... 31
4.2 CUSUM a tool for detecting persistent deviations from a background mean ... 33
4.3 The development of HCR’s for “data poor” stocks ... 34
5 References (includes both citations and useful references)... 37
Annex 1: List of participants... 39
Annex 2: Agenda ... 43
Annex 3: Working documents ... 44
Annex 4: Recommendations ... 106
Executive summary
WKFRAME-2 met for 4 days in January to provide further technical guidelines to as- sist ICES expert groups in the implementation of the ICES MSY framework for advice which was introduced in 2010. The workshop was attended by scientists from the ICES community, stakeholders from the fishing industry and environmental interest groups. This year a particular focus of the group was trying to develop technical solu- tions to issues which proved problematic with the advice formulation in 2010. For this reason the meeting was overlapped with WGCHAIRS, which provided an im- portant input to the first term of reference, “to evaluate the implementation of the advice and identify areas where further development is required”. Apart from tech- nical issues related to model fits of SRR, a major issue which arose last year was the confusion of what ICES was advising in relation to sustainable harvest rates, in situa- tions either where stocks were reproductively impaired, or where there were techni- cal differences between fishing at a defined Fmsy and according to an accepted management plan. The latter issue is now clarified and it is ICES policy to have a hi- erarchical approach to advice, where agreed management plans will be the primary consideration. In the situation where stocks are at risk of productivity impairment, the approach was to provide several options on the slope of the advised F rule where SSB is below MSYBtrigger. The decision on the implementation of any of these options is the responsibility of ACOM. WKFRAME suggests that adopting a singular ap- proach will make it possible to give advice using the ICES MSY framework which is consistent with both the PA and MSY approaches. Several technical issues arose in 2010 also in regard to the calculation of advised exploitation under the transition schemes. The approach where SSB is above MSYBtrigger, is relatively straightforward, and there is a suggestion to calculate the transition from F2010 and follow 5 equal steps to the Ftarget. In the case where SSB is or falls below MSYBtrigger during the transi- tion, the situation is more complicated. To help clarify the options, WKFRAME has produced generic equations which cover the various options and suggests that ACOM choose one of these options with a caution against the mechanistic application of whatever rule is chosen. Most of these issues fall under the second ToR “on [the basis of issues arising in 2010] to further develop the MSY approach” and the details are provided in section 2 of the report. Finally there was a ToR to “further develop the MSY approach to be applied in cases where no analytical assessment is available”.
The first issue here was to address the semantic concern where the labelling of advice as maximum sustainable yield, when it was based on relatively imprecise determina- tions of whether overfishing (in relation to MSY) is occurring, caused problems. A suggested solution is to label this advice as sustainable yield advice. A clarification of the guidelines from WKFRAME I, with some references to appropriate methodolo- gies is provided in section 4. In order to assist EG’s to provide the basis for determi- nations required by the ADG’s in drafting advice in situations where there is no forecast, a flow chart is provided.
1 Introduction
The first WKFRAME report (ICES 2010d) dealt with guidelines on robust approaches to estimating Fmsy, and guidelines for the application of the ICES MSY framework (ICES 2010b) to stocks where there was no short term forecast. This report focuses on technical solutions to implementation issues which arose during the application of the framework to provide fisheries advice in 2010. Some of the issues arose due to confusion in the multiple aspects of advice, i.e. advice in relation to MSY advice in relation to PA, and advice in relation to management plans. In other cases contro- versy emerged in what are essentially policy decisions on transition options to bring fishing mortality in line with targets by 2015. The issue of confusion due to multiple aspects to the 2010 advice was echoed by consumers of the advice. In response ICES will introduce a hierarchy in the 2011 advice whereby in the situation where a long term agreed management plan1 exists, this will have primacy for advice provision by ICES. Otherwise the advice will be given on the basis of the framework. The discon- nect between the PA and the MSY approaches in the 2010 framework also caused confusion and controversy. This report deals with the issue by suggesting a unified approach to the provision of exploitation advice, bearing in mind that fish stocks need to have full reproductive potential in order to deliver maximum yields at Fmsy. This logic is no different than that used to construct the ICES MSY framework, and any further developments proposed to the framework in this document are ulti- mately policy decisions to be made by ACOM. The logic for the approach taken in this report is outlined below (see also Section 3).
Stocks fished at Fmsy should fluctuate around a biomass which can provide maximum yield. In order to prevent against a condition of biomasses lower than this expected range, the ICES MSY framework uses the concept of a trigger point MSYBtrigger (in much the same was as any HCR), which simply triggers action of reducing the ex- ploitation from FMSY under the condition where the biomass moves out of the ex- pected range. MSYBtrigger is a biomass point which is expected with a low probability in a fully productive stock which is fished at Fmsy. In 2011 for those stocks which ICES gives advice under management plans, it is anticipated that this will continue to be the basis for the advice. For those stocks exploited near Fmsy, transition is expected to be relatively straightforward, and no additional considerations should be needed. In the case of some stocks for which ICES gives advice the exploitation is currently well above Fmsy and the biomass is at risk of, or is at a level, where recruitment impairment could be occurring. Under this condition a rebuilding is required before fishing at Fmsy can give maximum yield. Under the precautionary approach the condition of SSB>Bpa is a requirement in order to ensure that there is a reduced risk of productivity impairment. In the case of stocks which are in this condition the biomass Bpa can func- tion as an operational point below which fishing mortality is reduced at some rate to allow rebuilding to a magnitude where there is a low risk that the stock cannot pro- vide maximum yield under fishing at Fmsy. The use of Bpa in the ICES MSY framework in such a fashion is subtly different from its function in the PA, where fishing mortal- ity is only adjusted such that there is a neutral risk of being below the point. Here Bpa is simply being used as a trigger to reduce fishing mortality from Fmsy which is being applied as a target. The use of Bpa as an operational trigger point was suggested in 2010, even though it was considered that ultimately MSYBtrigger would correspond to a
1agreed management plan means endorsed by ICES and implemented by a competent authority
lower percentile of the biomass distribution under the condition of fishing at Fmsy. It is envisaged therefore that Bpa can act as a trigger point for stocks which require re- building. The rate of decrease of F below the point where recruitment is at risk of (Bpa) and becomes impaired (Blim), is a matter of risk tolerance and choice on time taken for recovery. Under the PA approach ICES has previously stated that its advice is risk averse to Blim and in the circumstance where productivity maybe impaired, its advice is given to affect a safe and rapid recovery. The choice of the rate of decrease in F be- low the MSYBtrigger should be consistent with this in order for a unified framework for advice under MSY and PA considerations to be appropriate.
The ToR’s of WKFRAME II are given below and dealt with as follows: Section 2 of the report deals with ToR a, Section 3 deals with ToR b, and section 4 deals with ToR c WKFRAME II Terms of reference
a ) Evaluate the implementation of the MSY approach (reference points, framework and transition) in the 2010 ICES advice and specifically identi- fying such areas where further development is required
b ) On this basis, further develop the MSY approach including:
i. Improved guidelines for reference point setting including FMSY, Btrigger and Blim which might be a consideration at low stock size;
ii. Implementation guidelines for the MSY framework including greater speci- ficity on how fishing mortality should change at low spawning biomasses;
iii. Put forward options for transition rules from 2011 onwards
c ) Further develop the MSY approach to be applied in situations where no analytical assessment is available.
Working documents presented at the meeting are included in Annex 3 at the end of this report.
2 Evaluation of the implementation of the MSY approach in the 2010 advice.
2.1 Overview of MSY reference points already defined
The relationships between F reference points from 2010 assessments were examined from data in “FishData 2010.xls”
Fig. 2.1.1 Scatter plot of Fpa vs Fmsy estimates from the 2010 assessments.
While the best fit to the scatter of Fpa against Fmsy (Fig. 2.1.1) indicates that Fpa is on average 1.6x Fmsy, there are a number of stocks for which Fmsy is estimated to be equal to or higher than the estimated Fpa (data points on or below the 1:1 line). These in- stances (sol-kask, her-irlw, her-noss, her-2532-gor, her47d3, her-noss, mac-nea) need to be investigated before the Fmsy points are confirmed as correct for use in the advice framework.
Fig. 2.1.2. Scatter plot of Flim vs Fmsy estimates from the 2010 assessments.
y = 1.5735x R2 = -0.065
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80
Fmsy
Fpa
y = 2.3785x R2 = -0.2277
0.00 0.20 0.40 0.60 0.80 1.00 1.20
0.00 0.20 0.40 0.60 0.80 1.00 1.20
Fmsy
Flim
On average, Flim is about 2.4x Fmsy (Fig. 2.1.2) which suggests a reassuring distance between these two reference points for most stocks, although a couple of data points fall close to the 1:1 line suggesting potential problems (see bootstrap analysis below).
Fig. 2.1.3. Scatter plot of Flim vs Fpa estimates from the 2010 assessments.
Flim is about 1.4xFpa (Fig. 2.1.3) which is not surprising since this is the default factor used in computations for many ICES assessments.
The relationships between MSY reference points, when the S-R model is of the hockey-stick type, were examined in a non-parametric (model conditioned) bootstrap analyses in R by Mesnil (2011). A thousand S-R parameter pairs were generated from the baseline fit of S-R for Norwegian Spring Spawning Herring and North Sea Cod by bootstrapping, which is assumed to preserve the sign (if not the magnitude) of co- variance. Hockey-stick S-R models were fitted to each bootstrap S-R samples and fed to the MSY calculation routine, yielding distributions of MSY reference points. Re- sults indicate that if Hockey-Stick is a valid representation of the actual S-R relation- ship (a big IF), Fmsy will most often be ≡ Fmax, hence independent of the S-R parameterisation. In the case where the YPR curve peaks at very high F, Fmsy coin- cides with Fcrash (i.e. close to Flim), and Bmsy coincides with the breakpoint in the Hockey Stick, i.e. Blim (see Figures of working document in Annex). This will create problems in applying the MSY framework for some stocks and suggests that me- chanical application of default methods should be avoided.
2.2 A note on the implied function of MSYBtrigger
In the 2010 report WKFRAME discussed the role of Btrigger and indicated it should be selected as a biomass that is encountered with low probability if Fmsy is implemented.
It was stated that if the SSB is below this level it is (by definition) out of expected range, and thus a suitable trigger to initiate action. Also WKFRAME considered that although Bpa is proposed as a default trigger biomass in the ICES MSY framework, it is not a logical candidate in the long term as it is derived from an error model basis around Blim, whereas under MSY exploitation is should be a property of the expected distribution of SSB. Thus Btrigger in the longer term should form two roles, a trigger for management action, first a need to reduce catches because SSB is outside the ex- pected range, but second it should also indicate a need to check that the basis for the stock dynamics is still valid. However, in practice by concentrating on SSB targets we
y = 1.4417x R2 = 0.8825
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80
0.00 0.20 0.40 0.60 0.80 1.00 1.20
Fpa
Flim
should not neglect recruitment. In situations where a recruit index is available or the assessment gives a relatively reliable estimate of recruiting year classes, a series of unexpected recruitment may be the first indication that the biological assumptions may no longer be appropriate. Thus there may be good evidence for questioning the applicability of the assumed dynamics a few years before the SSB declines, this was the case for NS herring, where the dynamics observed from 83-2001 had been consis- tent but from 2002 an atypical sequence of low recruitment was observed. It was possible to observe that the dynamics were atypical two or three years before the biomass reached levels that were of concern. In the evaluations of NS plaice and sole multi-annual plans it was noted that a sequence of reduced recruitment would give a much more timely indication of impending problems than waiting for SSB to decline.
EG should be encouraged to be aware of the assumptions underlying MSY exploita- tion for a stock and to look intelligently all indicators to confirm that the dynamic remain as expected. In such circumstances waiting for biomass alone to trigger a change in advice would not be making best use of available information.
2.3 Feedback from ICES EG’s experiences in 2010 Arctic Fisheries Working Group (AFWG) Feedback
The AFWG stocks can for the purpose of calculating MSY reference points be divided into 4 groups:
1 ) Stocks for which there is an accepted analytical assessment and an agreed HCR: NEA cod, haddock, saithe
2 ) A short-lived stock with survey-based assessment and an agreed HCR, and
for which a single-species MSY is meaningless since predation from cod and other predators is much larger than the fishery: Barents Sea capelin.
3 ) Stocks for which there are catch-at-age data and reasonable confidence in age readings, but no accepted assessment: Coastal cod, S. marinus, S. men- tella
4 ) Stocks for which the age reading methodology is under revision (Age Reading Workshop to be held in 2011): Greenland halibut
For the stocks under 1), harvest control rules have been evaluated using long-term stochastic simulations based on biological models with density-dependent growth and maturation. There are agreed management plans based on these harvest control rules, and the target F in the harvest control rules are in the range of F values associ- ated with high long-term yield. The biological models will be revised following benchmark evaluation of the respective stocks, and MSY calculations will then be updated accordingly. As for many other stocks, the yield curve is rather flat at the top, giving a wide range of F values which can be associated with high long-term yield. For NEA cod and haddock, the managers (Joint Norwegian-Russian Fisheries Commission) in 2010 agreed that the current HCRs should stay unchanged for 5 years and then be evaluated.
For Barents Sea capelin, the agreed HCR is that with 95% probability, at least 200 000 tonnes (Blim) should be allowed to spawn. There is no Bpa and no F-based reference points. MSY has been investigated by Tjelmeland (2005), using the multispecies model Bifrost. He found that the BMSY reference point of capelin depends markedly on
the harvesting strategy chosen for Northeast Arctic cod and Norwegian Spring- Spawning herring, which both have strong biological interactions with capelin. Thus, calculating a single-species MSY for capelin is meaningless. The capelin MSY could be calculated given the agreed HCRs for cod and herring, and one could then investi- gate whether the MSY for capelin would change considerably if the harvesting strate- gies for cod and herring vary e.g. within the intervals corresponding to yields > 80%
of the MSY for herring and cod.
For the stocks mentioned under 3 and 4, there are no F or SSB based harvest control rules, and better biological models are needed in order to calculate reference points.
WKFRAME comment: The approach taken by the AFWG is consistent with guide- lines produced on the implementation of the MSY framework by ICES last year. The introduction of an hierarchical approach to advice, giving primacy to implemented management plans, should solve the issues in relation to category 1 stocks above. For category 4 stocks there is no immediate prospect of producing proxies for MSY refer- ence points until the growth issues are sorted. For Category 3 stocks WKFRAME would suggest that progress could be made in identifying exploitation proxies (See WKFRAME I section 2.2)
North Western Working Group (NWWG) Feedback
During discussions in the group it was noted that simulations show that identifying a single FMSY value is almost an impossible task. Effectively one can only identify a range of fishing mortalities that all would conform to the MSY concept. Hence, the NWWG considers that the only appropriate method to evaluate the MSY principle is by evaluating catch rules in a stochastic simulation framework that takes into account both natural and assessment noise (as has been done e.g. for the Icelandic cod and saithe and to some extent the Icelandic haddock). The resulting fishing mortalities that lead to optimum yield obtained within such a framework are not analogous to the FMSY proxies obtained from applying the short-cut methods suggested e.g. in WKFRAME. On a similar note the current setup of the ICES advice summary sheet stock table can send out a wrong signal where when it states that the MSY reference points have not been defined but the stock is managed according to a HCR that has been evaluated to conform with the MSY-approach but that HCR does not have ex- plicit Btrigger and FMSY points as is the case for cod in Va.
WKFRAME comment: The introduction of an hierarchical approach to advice, giving primacy to implemented management plans, should solve the issues in relation to many of the stocks dealt with by the NWWG.
Working Group on Elasmobranch Fisheries (WGEF) Feedback
The Working Group on Elasmobranch Fishes attempted to apply the framework to each of its stocks. Certain issues arose with mixed species stocks, such as “Demersal Elasmobranchs in the Celtic Seas” where several species of ray are landed without species identification. Comparative species vulnerability was taken into account when formulating advice.
In addition there are certain species at low stock levels and with very low fecundity, where FMSY is not an achievable target in the short term. Such stocks include deepwa- ter shark species.
In general, the Guidelines for stocks where no analytical assessment is available, were fol- lowed.
WKFRAME comment: The approach suggested this report is compatible with the WGEF decision to advise on Precautionary considerations where the stock productiv- ity was impaired. This short term prioritization of the PA is rationalized in order to have any prospect of achieving MSY.
Herring Assessment Working Group (HAWG) Feedback.
In 2010, HAWG met before WKFRAME, however good progress was made in devel- oping the new ICES FMSY framework. Medium term simulations were conducted using the HCS10 software. This is a medium term projection program designed for exploring harvest control rules, without doing a full assessment as part of the annual simulation loop. The program is a recently revised and updated version of the HCM/HCS software that has been used for evaluation of management plans in the past (mackerel, blue whiting and Celtic Sea herring). It has an age based population model in the background with stochastic recruitments but fixed weights and maturi- ties, an 'observation' (assessment) model that produces a noisy basis for management decisions, a management rule module with various options, and an implementation module that translates management decisions into real removals, again with noise.
Yield and biomass per recruit is calculated as a by-product. The program was run over 50 years with a range of fixed fishing mortalities as the management decision rule, with no modifications. The risk presented is the fraction of the iteration trajecto- ries where the SSB is below Blim in year 50. Yield and biomass per recruit and F0.1 are produced as a by-product.
The approach taken by HAWG was later incorporated in the work of WGFRAME.
HAWG interpreted FMSY as a value of F that is expected to lead to a near maximum yield in the long term. For most stocks, there will be a lower bound where long term yield is lost because of low exploitation and an upper bound where there is an in- creasing risk of recruitment impairment. Within that range, there may sometimes be a distinct maximum, depending on selection at age, growth rate and natural mortality.
The pattern may be modified if growth and maturity are density dependent, or if the natural mortality is sensitive to multispecies effects.
For most herring stocks, which typically are lightly exploited at small size and young age, there is no distinct maximum. Hence, the highest long term yield may be ex- pected at a fishing mortality which is close to that leading to recruitment failure. The lower bound may be represented by F0.1, but in some cases F0.1 may be higher than the mortality leading to impaired recruitment. Hence, the most rational target fishing mortality may be one where the loss is small, and which is safely away from the re- gion where the recruitment may be impaired.
Of the six stocks considered by HAWG, 2 have existing management plans, and 3 more had management plans under development. None of these plans is inconsistent with FMSY, though some do not have the same HCR as the generic ICES FMSY HCR.
HAWG outlined the region of fishing mortalities associated with a near maximum long term yield by calculating yield per recruit combined with a stock-recruit rela- tionship. The effect of random variation in the recruitment in a stochastic equilibrium was evaluated, but not the uncertainties in assessment and implementation, nor variation in weights, maturity or selection.
Yield per recruit is sensitive to natural mortality, growth rate and selection at age, and assumes that all these are independent of F-level and stock size. This may not be true, and change in these factors may lead to a quite different perception of the shape and level of the yield per recruit curve, as well as the risk to stock collapse.
HAWG did not consider candidate values for a Btrigger have in great detail. HAWG considered that there is a range of biologically appropriate biomass triggers are pos- sible for each stock. The final choice will most likely be made based on management plan development. As such the trigger biomass has already been, or will be subject to evaluation by ICES. HAWG regards the development of management plans as the way forward to a rational utilisation of the resources, and is concerned that too strong an emphasis on specific values for FMSY or BMSY may hamper the development of good management plans.
The table outlines some values of F and SSB that may be a guidance to setting FMSYand Btrigger. The suggested values are suggestions only. Biomasses are in thousands of tonnes.
F range FMSY Btrigger Management plan
Stock L U Suggested 10th %ile SSB
at suggested FMSY
Suggested Btrigger F
North Sea herring 0.15 0.25 0.25 (MP) MP 800 to
1,500 0.25 (@ high SSB)
Western Baltic 0.22 0.3 0.25 170 UD* UD*
Via (North) 0.17 0.35 0.25 (MP) 85 MP 62.5 and
75 0.25 (@ high SSB)
Via (South) & VIIb,c** 0.2 0.28 0.25? 95 UD** UD**
Celtic Sea 0.18 0.3 0.25? 50 UD*** UD***
Irish Sea NA NA NA NA NA NA NA
* As per simulation work in support of management plan development, underway in Jakfish Pro- ject.
** No analytical assessment available to estimate a TAC for a given F. Stock recruit information taken from converged VPA, as per simulation work conducted by Irish Marine Institute in 2010, in support of management plan development. Other inputs from sVPA using a terminal F of 0.5, considered the most informative exploratory assessment (Chapter 6).
*** As per simulation work conducted by Irish Marine Institute in 2010, in support of management plan development in conjunction with stakeholders’ committee in Ireland.
MP: As per existing management plan UD: Management plan under development
Setting a FMSY for a stock mixing with other stocks as the WBSS-NSAS complex in Division IIIa and adjacent areas proved rather difficult as reaching FMSY for all stocks involved in the mixing is impossible in practise. Thus for the WBSS, the WKWATSUP decided on a TAC setting procedure only acknowledging the weaker stock in the mix. Thus, the TAC should first be set for the WBSS according to the FMSY or FMSY transition framework for WBSS alone. If the NSAS is greatly im- pacted by management of the WBSS, this rule needs to be re-evaluated. Following this, the fraction taken in the Eastern part of the North Sea (parts of Sub Divisions IVb and IVaE) should be subtracted from the total TAC for the WBSS before sharing the TAC between Division IIIa and Subdivisions 22-24. Subsequently the best estimates of the proportions of the NSAS and WBSS in the catch by fleet should be used to cal- culate the combined catch options in compliance with the targeted catch for WBSS.
WKFRAME comment: The introduction of an hierarchical approach to advice, giving primacy to implemented management plans, should solve any issues in relation to advice provision where there is a management plan. The mixed fisheries issue in rela- tion to the WBSS-NSAS complex is not solvable from a purely scientific perspective.
The decision to provide TAC advice on the basis of not overexploiting the weakest
stock (in the mixed fishery) is a policy decision, and it should remain transparent as such.
Working Group on Widely Distributed stocks (WGWIDE) feedback
WGWIDE 2010 was held over a shorter time than previous years and stock co- ordinators generally felt that the requirement to calculate new reference points (i.e.
FMSY and Btrigger) posed a hindrance to the main purpose of assessing the stocks. There was uncertainty over what exactly Btrigger was supposed to represent and how this should be determined. The ADMB plot MSY methodology was used for most of the stocks to examine potential FMSY values. Despite all stocks (with the exception of North Sea horse mackerel) being ‘information-rich’, there were still difficulties in es- tablishing FMSY levels in the traditional equilibrium manner. This was mainly down to what were perceived as poor stock-recruit fits and uncertainty over the appropriate S-R functions. In addition numerous stocks had very flat-topped YPR curves on which Fmax (a fallback proxy for FMSY) was poorly defined.
For NEA mackerel and blue whiting, both of which have recently had management strategies evaluated, the fallback position was put forward the simulation tested tar- get F value as a candidate for sustainable management yielding high long term catches. For NSS herring, a simple stochastic simulation evaluation using a Beverton- Holt S-R function and measurement error was used to derive a safe, high yield F value which was proposed as FMSY for this stock. Stochastic yield per recruit analysis (using plotMSY) was used for western horse mackerel, where F0.1 was proposed as a FMSY proxy given the close proximity of Fmax to Fcrash. No values were proposed for North Sea horse mackerel (no assessment, only catch values). The integrity of this stock unit (whether it is a closed stock or not) most probably would prevent a viable FMSY value being determined.
There was substantial confusion over how the transition approach should be imple- mented, particularly in stocks where B < Btrigger (i.e. blue whiting). There were also problems with the acceptability of advice arising from this transition scheme, particu- larly when there was a large discrepancy between PA approach/management plan and the MSY advice. The gradual step down towards FMSY may lead to advice that appears as unsustainable. This perception may be changed by presenting both what the advice would be purely according to FMSY as well as the intermediate FMSY transi- tion advice.
The frequency of reference point revisions was raised as an issue during the ICES- Pelagic RAC meeting on the ecosystem approach to fisheries. Stakeholders expressed concern over the view that values proposed now are set in stone. They thought diffi- culty in revising values in the face of changes in stock productivity (e.g. regime shifts) or changes in the fishery was a cause for concern. It was felt that guidelines on when and how to revise FMSY reference points would be useful.
In Summary:
Stochastic simulation analyses were preferred to equilibrium analyses because of poorly defined stock-recruit relationships or flat-topped YPR curves.
The rationale for Btrigger was ill-defined.
There were issues both in applying and the advice arising from the transition scheme.
WKFRAME comments: The issue of calculating reference points providing “a hin- drance to the main purpose of assessing the stocks” could be helped if the ToR for EG
related to conducting the analysis required to provide an exploitation advice following ICES guidelines rather than simply update the assessment. The introduction of an hierarchical approach to advice, giving primacy to implemented management plans, should solve any issues in relation to advice provision where there is a management plan. In rela- tion to stakeholder feedback to WGWIDE on the frequency of reference point revision WKFRAME I suggested that Fmsy targets could be set provisionally and updated when more data and/or more complete analyses had been carried out, in this context Fmsy reference points could be annually updated until such time as these analyses are complete. With regard to the identification of regime shifts and how to advise when there are shifts in productivity, this issue is discussed in section 2.3.2 (below).
Working Group on North Sea and Skaggerrak (WGNSSK) Feedback
Four different approaches were developed by WGNSSK, largely developed around ICES WKFRAME and further used and developed during the WGNSSK meeting. The first three deal with stocks for which age-based information exist, and present many similarities in their standard combinations of YPR, SRR and SPR relationships. The fourth one is an approach specifically developed for Nephrops stocks ahead of the WG meeting. 1) A suite of programs (built in AD model builder) was successfully tested and used for a number of stocks during WGNSSK meeting, and served as the primary tool by the WGNSSK for providing final Fmsy estimates. 2) A number of R scripts using the FLR framework (www.flr-project.org) were developed ahead and during ICES WKFRAME 2010 (Case Studies 3 and 6). These scripts were later merged into a single generic R-FLR program (Finding Fmsy with FLR_v4.r), in order to explore and compare various methods for estimating Fmsy using a single FLStock object as input. An alternative R script was also developed around ICES WKFRAME 2010 (Case Study 5), using a analytical combination of fitted stock-recruit, yield-per-recruit and SSB-per-recruit curves. This script was used during WGNSSK for estimating Fmsy for the haddock stock, and is described in the corresponding section 13.7 in WGNSSK report. 4) The method developed for Nephrops is described under the WGCSE sec- tion below.
Summary: The MSY reference points estimates were found to be highly dependent of the underlying hypotheses. In a single-stock age-structured assessment context, the main problems encountered by the WGNSSK were:
• The usually very poor fit of the SRR estimates. In most cases, there is no evidence of any relationships;
• When this is the case, then different software (R/FLR versions, ADMB) may give very different parameters values, in particular for the break point in a Hockey-Stick relationship.
• Ricker-type SRR with a maximum estimated outside of the range of histor- ical observations should not be used as a basis for analyses
• The number of years used for averaging the weight-at-age and selectivity- at-age values can play a major role. ICES WKFRAME (2010) recommended taking the longest time-span where no significant trends are observed;
however, a number of software may take a three-years average as default value.
• Different values are usually obtained when estimating Fmsy using equili- brium equations (e.g. using the FLBRP package in FLR) or using stochastic projections
• Inclusion of multispecies considerations and density-dependent biological parameters can be an issue when considering that current parameters often correspond to periods of low abundance, and these may be quite inappro- priate in future high abundance levels.
WKFRAME II comment: WGNSSK spent considerable time and effort exploring and developing methods and software to estimate Fmsy targets for the stocks covered by the WG. The problems encountered which are due to lack of correspondence between the data and assumed model, where they are not simply due to a short time series or lack of dynamic range, may simply be a property of the data and there is no analyti- cal solution for that situation. In such situations expert judgment should be applied in order to determine the most appropriate outcome. Software defaulting to inappro- priate time ranges is a technical issue which requires a technical fix of disabling au- tomatic defaults and manually setting or hard coding the appropriate time range.
Differences in Fmsy estimates between equilibrium and stochastic approaches can be expected. The best practice advocated by WKFRAME would be to use stochastic si- mulations; the approach is elaborated in Section 3.1. Dealing with density dependent biological parameters can be problematic, but a rational approach would be to use expert judgment and at least include as a source of uncertainty.
Working Group on Celtic Seas Ecosystem (WGCSE)/Working Group on Hake Monk and Megrim (WGHMM) Feedback
WGCSE used ADMB to explore the S-R, fishery selection, and growth potential data for fin-fish stock where assessment data were available. Based on an analysis of the uncertainty, the idea was that the most plausible S-R relationship would be used for the estimation of FMSY. However in many cases more than one plausible S-R function existed and the FMSY estimates differed in the absolute values depending on the form of the S-R model used. Where this was the case WGCSE concluded that no definitive value of FMSY could be defined and the range was provided to the ADG.
For stocks with no assessment the Working Group performed qualitative evaluations and suggested, were possible, the current stock status in relation to FMSY. Considera- tions were given to detail such as trends indicative of stock status, age structure, dis- card rates, status of other exploited stocks in a mixed fishery. Directional advice were proposed and major problems such as high discard rates were highlighted and possi- ble management measures suggested to reduce discards with the presumption that it will increase future yield.
For Nephrops stocks, given the differences in fisheries and ecology it is inevitable that estimates of the exploitation rate leading to long-term MSY will vary between the FUs. Given this the approach taken by WGCSE depended on the data available. For those stocks with a TV survey, the Harvest Rates (removals divided by abundance as estimated by the TV survey) associated with fishing at F0.1 and Fmax were estimated at the 2009 benchmark meeting WKNEPH. In response to the recommendations of WKFRAME, estimates of F35%SpR and the corresponding Harvest Rate were also de- termined and these estimates typically lay between the estimates of F0.1 and Fmax. Sug- gestions for a TV-abundance based proxy for Btrigger were made on the basis of the lowest observed TV-abundance (median survey value) unless the stock has shown signs of stress at a higher TV-abundance in which case this value was proposed Btrig-
ger. The remaining challenge is determining which Fmsy proxy is appropriate for which stock and this becomes an exercise in expert judgment based upon knowledge of the fishery and the ecosystem. In order to assist communication of the expert judgement
process the following bullet list is suggested as a standard checklist for describing the rationale behind the choice of a particular Fmsy.
• Describe the absolute density. Is it high (i.e. >1 per m2), medium (i.e. 1.0–0.2 per m2) or low (i.e. <0.2 per m2)
• Variability in density. Is there large interannual variability, spatial complexity?
• Understanding of biological parameters. Is the growth rate particularly fast or slow, high or low estimates of natural mortality?
• Fishery timing and operation. Is there a strong seasonal pattern leading to different exploitation rates on the sexes, does this pattern vary much between years?
• Observed Harvest Rate or landings compared to stock status. Is the harvest rate consistently around or above Fmax? Have landings been stable? Have the indicators of stock status shown signs of difficulty?
Accompanying this text should be a table listing the Fmsy proxies Fmax, F35%SpR and F0.1 for males and females, the Harvest Rates they correspond to along with the implied
%spawner-per-recruit for males and females.
Summary:
For stocks were a possible range of FMSY were proposed, a single value was put for- ward in the final advice even though the WGCSE proposed a range. There was no real consistency in the choice of FMSY across stocks within management areas and in some cases FMSY was set by analogy to other stocks of similar species.
Some issues arose from the implementation of the ICES transition scheme where stock biomass was below Blim, but the MSY advice was for increased catch option above the current TAC. This was purely a function of the rigid implementation of the transition framework.
There are numerous stocks in the Celtic Sea ecoregion with no assessment and most of the issues and problems were associated with these stocks. The general approach for these stocks was to avoid generating rash FMSY values, but rather focus on provid- ing directional advice, i.e., proposed management measures to get F closer to FMSY for stocks that are likely to exploited well above FMSY. With the information available for these stocks it would be possible to generate specific FMSY targets, but with no esti- mate of current F the setting of quantitative advice remains problematic. Quantitative advice is relatively simple if the stock is exploited well above FMSY, but advice for stocks that are exploited close to FMSY remains challenging.
WKFRAME comments: The approach taken by WGCSE and WGNSSK is consistent with guidelines suggested by WKFRAME I. Some issues surrounding model fits to SRR are touched on in this report (Section 2.1. & 3.1). Many of the problems relate to short time series and poor correspondence between the assumed model and the data.
Where this problem may be due to limited data an alternative generic species ap- proach outlined in this report section 2.3.1 (below) which could provide a way for- ward. The final problem raised by WGCSE is in relation to providing quantitative advice for stocks which may be exploited close to FMSY, using the framework outlined in the introduction to the advice last year. This problem arises when the advice is only directional in relation to the catch on the basis of whether the stock is overex- ploited or not. The determination of whether the stock is over exploited or not deter- mines the basis for the advice and this determination may rely on expert judgement to a greater or lesser extent. There are currently no guidelines to prescribe this expert
judgement, and WGCSE indicate that they can foresee consistency problems arising where no guidelines exist. It may be that these difficulties relate to scientific argu- ment and fishery specific issues, for which there will always have to be an accommo- dation in any rational basis for fisheries advice.
WGDEEP Feedback
WGDEEP had very little time to assimilate the guidance on the implementation of the ICES MSY concept from WKFRAME, however a number of recommended ap- proaches for data-poor stock were explored, mainly using southern blue ling (Vb,VI and VII) as a case study. These were: Depletion corrected average catch ((MacCall, 2009); Catch curve analysis; and Productivity-susceptibility Analysis (PSA). Within the time constraints of the 2010 meeting, it was not possible to develop any of these approaches to the point where they could be used as a basis for advice but the Work- ing Group recommended that further work on developing MSY reference points should be added to the ToR for 2011. Because MSY proxies could not be established for any stocks, draft advice for all stocks in 2010 was given relative to the precaution- ary framework as in previous years.
WKFRAME comment: The situation where there is no short term forecast arises fre- quently for the stocks covered by WGDEEP. In such cases making a determination of whether the stock is overfished or not and establishing the trend (if any) in abun- dance become the primary objective of the ICES MSY framework. Where size and growth data exist and there are informative abundance indices (i.e. standardised), the approach outlined by WKFRAME I and further elaborated here in Section 4 can be followed. This may involve some investigation (e.g. sensitivity analyses), but current and sustainable exploitation proxies can be derived easily to give a guidance for ad- vice.
2.4 Summary of areas identified for further development 2.4.1 Sole meta analysis
Meta-analysis, applying generic methods within a species group, has benefits in terms of providing more stable estimates of FMSY reference points and estimates for individual stocks which by themselves may otherwise have insufficient data. Current ICES advice for seven sole stocks for which there is sufficient data gives Fmsy esti- mates which vary from 0.16 to 0.38, but these are specified for differing age ranges so are not directly comparable. Bpa is estimated for six of these seven stocks but Blim is available for only two. Slope at the origin estimates for HS models for the seven stocks vary from 1.32, to 4.45 largely due to observed exploitation rates. An evalua- tion by Simmonds (2011) evaluates the use of a consistent framework to compare Fmsy
targets and the sensitivity of long term yield to the choice of this value across these stocks.
Recruitment is modelled though stochastic multiple model based simulation for 1000 constructed “populations” for each stock by randomly sampling with replacement selection at age in the fishery, weights at age in the catch and weights at age in the stock for the period 2001-2009. S-R models were fitted in a Bayesian framework under the assumption that sole has generic exploitation form which can be scaled to the car- rying capacity for each stock, but the resilience, or slope (R/SSB) to the origin, is con- sistent across stocks assuming each stock retains its own estimated or assumed growth and maturation. Three S-R models were applied, Hockey-stick, Ricker and Beverton-Holt with each model formulated so that the A parameter defined the slope
to the origin which was assumed to be common for all stocks, whereas the B parame- ter related to density dependence was assumed to be independent.
The probability of each model type is selected for the set using the method described in Simmonds et al (2011).
Results show that changing from only HS to multiple models has a minor impact on the estimate of Fmsy or the F that maximizes mean catch, though the range of estimates is reduced. There is a small reduction in range through standardization of F bar over the same ages (3-8). A slightly larger reduction using mean selection at age in the fishery. The largest contribution comes from the different growth of the sole in differ- ent areas. In most cases if exploitation is at Fmsy there is less than 5% probability of SSB being less than Blim, except for NS sole. For NS sole measurement error will have some impact on the results whereas for other stocks the influence is limited.
The combined analysis (Table 2.4.1) shows that the range of F within which catch is within 5% of maximum catch is substantial. Only for Skagerrak-Kattegat sole is the target near the lower bound. For most stocks an F target (ages 3-8) of 0.25 is a good choice. The new Blim and Bpa values are an important change, they are coherent with the simulations and contribute to the impression of safe exploitation. However they need to be considered carefully before acceptance, this further analysis would be most appropriately conducted by individuals familiar with the stocks. The higher exploitation rates for Irish Sea and Skagerrak are conditional on the growth and this need to be verified, particularly for Skagerrak-Kattegat.
Table 2.4.1. Comparison of Median Fmsy estimates from the combined analysis in this study and those from ICES stock assessments.
Stock Mean age -5% Catch lower
F for Max
Catch Median
Fmsy -5% Catch
upper
Ages for ICES Fmsy
ICES Fmsy 3-8
BoB 3-8 0.15 0.25 0.24 0.4 3-6 0.27
CS 3-8 0.15 0.25 0.22 0.35 4-8 0.31
EC 3-8 0.15 0.25 0.27 0.35 3-9 0.29
IRS 3-8 0.25 0.35 0.4 0.55 4-7 0.15
NS 3-8 0.2 0.25 0.35 0.4 2-6 0.24
S-K 3-8 0.35 0.4 0.43 0.65 4-8 0.36
WC 3-8 0.15 0.25 0.25 0.4 3-9 0.27
2.4.2 Detecting Regime shifts
Guidelines for determining ecosystem regime shifts
The issue of ecosystem regime shift (RS) will here only be considered in the context of S-R analysis within analysis of biological reference points for advice and manage- ment. Thus, changes in weight at age due to density dependence, accounted-for changes in natural mortality due to changes in predator stock biomass, etc., will not be considered.
Philosophically it might be fruitful to consider the following question: How can we sensibly identify ecosystem parameters of importance for a particular fish stock re- garding RS, when we have no clue on which parameters that are influencing recruit- ment variability (except SSB) - are we introducing an inconsistency in our system by considering RS?
This issue of RS is related to the classic dilemma between having a long time series of data and a large dynamic range, versus considering a (fairly) constant ecosystem re- gime existing only for a shorter time. Due to the large variability of recruitment a time series of say 20 years is a short time series in the context of estimating S-R pa- rameters.
Can individual years be regarded as a RS? Or is that better dealt with as noise? What about two years, three years etc? Is there a minimum length in terms of number of years for a regime?
It is important to realise that a regime shift does not have to be sudden, but can also be gradual.
It is also important to realise that the time series do not have to be continuous. If there is a temporal anomaly like the Gadoid Outburst for the North Sea, then it might or might not be appropriate to delete a time window and not all data points before the end of such an event.
RS can be a result of fisheries management, e.g. for the Baltic Sea the high F on cod has driven the stock to a low level and the sprat stock has increased simultaneously due to low predation from cod. Sprat in turn eat cod eggs and the cod S-R seems thus to be in a new Regime. Thus, theoretically fisheries management can in this case turn the regime back if wanted.
It is also worth considering that when a RS has been identified, is it then best to com- pletely ignore data related to the anomaly period or can some useful information be extracted from e.g. the S-R prior to the RS?
RSs seem to influence the fishing mortality Reference Points (RP) more than the bio- mass RP:
a ) Baltic cod S-R breakpoint constant over time because it is mainly due to cannibalism, however at times of good environmental conditions with high cod egg survival, F can be increased and still keep the biomass above the breakpoint;
b ) Multispecies modelling results often have shown that allowing biomass to increase under good environmental conditions decreases the productivity of other species by predation;
c ) “Pandalus goes up when cod stocks go down” in 17 out of 18 ecosystems considered by Worm&Myers (2003);
d ) “Nephrops goes up when cod stocks go down” in the several Nephrops stocks in the greater North Sea area and the Irish Sea;
e ) Density dependent growth and reproductive output per individual fish (as illustrated by liver index) of Northeast Arctic cod.
f ) Food depletion of Baltic sprat and herring by high stocks biomasses, result- ing in reduced growth.
WGIAB (ICES 2008a) and WKEFA (ICES 2007a) have given plenty of very useful points in relation to identifying RS and how to handle RS in scientific advice to man- agement.
Notes from WKEFA Report (2007)
WKEFA considered how change due to environmental factors can be included di- rectly in management advice. As ICES moves towards providing longer term advice in a rapidly changing environment there is a need to alter the way we consider the future and to provide advice that is both more robust and more adaptive to change.
WKEFA attempted to formulate generic solutions to identify, develop and evaluate procedures for improving fisheries management strategies and advice by including environmental information. Two types of variability were defined, stochastic stability and regime shift. Stochastic stability treats the short term as variable around a stable point and regime shift as long term different centres of stability. Both need to be taken into account in developing scientific advice.
For stocks in a relatively healthy state the dominant characteristic for consideration in a management advice context is the carrying capacity (recruitment at medium to high biomass), reflecting available long term yield. For those in a depleted or recovery phase, the productivity (rate of increase in recruitment with biomass at low biomass) will be the dominant factor. The importance of understanding how the environment influences these two aspects therefore depends not just on the stock but also its state.
Medium-term simulations should include evaluations under different environmental regimes to determine robustness to different plausible possibilities, rather than ex- pecting to optimise management under all conceivable options. It is unlikely that a single management strategy will be optimal under different regimes. On the basis of simulations WKEFA concluded that regime specific fishing mortality management strategies can be used as a tool for contending with decadal-scale climate or environ- mental variability. These management strategies outperformed constant fishing mor- talities management strategies by providing a balance between benefits (high yield) and trade-offs (fishery closures).
Simulations suggest that fishing mortality based management strategies are more robust to regime shifts than biomass related management strategies (Kell et al. 2005), because regime changes often result in changes in carrying capacity leading to differ- ent equilibrium biomass. The necessary time frame to detect regime shifts in fish communities depends on the life history, the age of recruitment to the fishery and the exploitation rate (MacCall 2002; King and McFarlane 2006). Shorter lived species with low age of recruitment and high exploitation rates require rapid detection of change.
Management for very short lived species normally involves rapid response to fluctu- ating recruitment, so such management regimes tend to have to respond to change quickly anyway, thus making them more adaptable under conditions of regime shift.
If the management regime for short lived species is not robust, changes in regimes will make the situation even worse (Polovina 2005). In contrast long lived species ex- ploited at a low rate and with older age of entry to the fishery allow for slower man- agement response (King and McFarlane 2006).
Notes from WGIAB (2008)
WGIAB investigated 7 sub ecosystems of the Baltic, covering besides the Central Bal- tic Sea, the Gulf of Riga, the Gulf of Finland and the Bothnian Sea additionally the Sound, the Bothnian Bay and a coastal site of the Swedish coast. Future analyses of
the Kattegat system are planned. Using multi-variate statistical analyses WGIAB demonstrated pronounced climate, fisheries and eutrophication related structural changes (i.e. regime shifts) in these regional Baltic ecosystems. About 30 different ecosystem parameters were analysed. The results showed that different selections of set of parameters gave wide variation in years of regime shifts and considering this large range of parameters were very important for determining reliable regime shifts.
Changing state:
In some cases environmental state changes are not directly reversible, although a change may cause a productivity change and may alter carrying capacity for a stock or stocks in a region, the reversal of that state change may not result in the subse- quent reversal of the state of the stocks. The situation for managers is perhaps differ- ent when a shift is from a more favourable to less favourable regime, however there are also issues when the shift reverses. In the first case it is necessary to adapt to the lower productivity. However, in the reverse case it is important that management actions in the low productivity period do not prevent the reversal from occurring. For example it may be necessary to maintain sufficient stock size in a low regime to allow that stock to recover, once the more favourable regime occurs. Thus understanding the system response is critical to determining the safe exploitation under low produc- tivity conditions.
2.4.3 The development of HCR’s for generic species types
In 2008 an STECF working group (STECF 2008) tested the performance of some ge- neric HCRs in a full MSE approach (De la Mare 1998, Punt and Donovan 2007) using FLR (Kell et al. 2007). Recently ICES (ICES 2010a) used the same simulation model and parameterization to test Annex IV HCRs (Anon 2010). The Operating Models were built based on two stocks with different life histories, Cod and Herring. For each of them two different stock recruitment relationships (SRR) and three different starting points were used. The SRRs had the same functional form for each stock, Ricker for Cod and Beverton and Holt for Herring, but they differed in their steep- ness. Steepness values of 0.75 and 0.9 were used to represent less productive and more productive stocks. The 3 starting points were relative to the exploitation level of the stocks, well managed stock (F < Fmsy & SSB > Bmsy), stock experiencing overfishing (F >Fmsy & SSB > Bpa) and overfished stock (F > Fmsy & SSB < Bpa). The observation error model introduced error only in a simulated CPUE using a lognormal distribution with a CV of 30%. In some of the scenarios a retrospective bias was introduced in the CPUE. Regarding reference points, F0.1 was used as Fmsy proxy and Fpa was defined as 2×F0.1 for the cod-like stock and 3×F0.1 for herring-like stock. Different multipliers were used for the two stocks because in the case of the Ricker (“cod”) stock recruitment relationship, the slope at the origin is less steep than for the Beverton and Holt (“her- ring”) formulation, and so FCrash occurs at a lower level of fishing mortality. Then Bpa
was calculated as the SSB at Fpa.
In STECF work a model free HCR based on CPUE and 2 HCR based on VPA (XSA) results were used. Among VPA based HCRs one used F0.1 as a target and the other the maximum between Fsq and F0.1. Besides both HCRs had common annual limita- tion in catch variation and more restrictive rules when the SSB fell below Bpa. The ICES work was focused in data-poor stocks and 4 different HCRs were tested. Two of them were fishing mortality based and used 3 years catch at age data to carry out a pseudo-cohort analysis and un-tuned VPA analysis. The other 2 were biomass based
and used CPUE as a proxy of biomass; one used a step rule to set the TAC and the other a linear transition rule.
The STECF model free HCR was found to be dysfunctional. Its effect was to increase fishing mortality leading to an initial increase in yields but this was quickly followed by accelerated increases in fishing mortality, reductions in SSB, yield and eventual stock collapse. The use of a constant multiplier on fishing effort (and hence fishing mortality) was obviously not sufficiently adapted to changes in observed CPUE lev- els. Implementing a variable (delta E) to prevent run-away increases in fishing mor- tality may correct that failure and provide time for the CPUE to change and the HCR to adapt to changing conditions within the fishery.
The F-based rules in Annex IV applied to severely limited data conditions performed exceptionally poorly in terms achieving their intended target of Fmsy. The biomass- based rule based on a step function, performed poorly because it responded to changes in total biomass when these changes were too small. A modification based on a linear transition rather than a step function, was more responsive to changes in total biomass and therefore performed better overall, in terms of achieving a stable SSB, when the biomass index was reliable in terms of trend. The objective of keeping SSB stable did not deal with the question of whether that SSB level was appropriate or will lead to optimal yields over time. Performance deteriorated for both biomass- based rules when retrospective bias was added to the survey.
The VPA based HCR which chose the maximum between Fsq and F0.1 often led to some rebuilding and recovery. However it often failed to improve situations where overfishing was occurring and even constituted a risk to well managed stocks. In these cases the rule either maintained fishing mortality at too high a level, preventing recovery, or it led to a gradual increase in fishing mortality leading to slow stock de- clines. The HCR could become stuck on relatively high fishing mortality rates that can harm or continue to harm stocks. This occurred because the Fsq was often too high to be sustainable. Finally the F0.1 based HCR maintained well-managed stocks and recovered stocks that had experienced overfishing or were being overfished. This re- covery occurred even in the face of a retrospective bias, although the improvements and level of rebuilding were often reduced. This HCR would often lead to a reduction in yields for the first few years after the introduction of management. Performance in terms of the development in yield and stock biomass for overfished stocks was im- proved and the risk to well-managed stocks was reduced considerably.
There is probably much to be learned by exploring generic analyses of both species productivity (see Annex 3.1) and HCR’s. In the former case (with careful circumspec- tion) realistic target F’s can be developed, where the individual stock data do not support such analyses, and in the latter case generic analyses can expose potential mismatches between the specifics of certain rule based approaches and the potential uncertainty and bias in the metrics of the stock response to the fishery. There is a fur- ther look at the development of HCR’s for “data poor” species in section 4.3
3 Further developments of the ICES MSY approach
3.1 Choice of methodology to derive Fmsy values.
Advice should be prepared based on targets that are derived from various methodol- ogy. The following list forms a hierarchy of sources of estimates of suitable targets.
WGs should select values that they consider are the most suitable for MSY exploita- tion and the following list provides guidance on which methods are expected to be preferred. Estimates derived from studies higher up the list are the preferred values.
i ) ICES endorsed multi-annual plan (not a target F but a complete plan in all its aspects)
ii ) Stochastic population model evaluation including errors (giving target F or agreed harvest rule)
Including assessment routines within a feedback framework with er- ror
Feedback framework using an error model (without assessment) Population model including varying biological parameters (annual or cohort effects) but excluding errors
Analytical yield per recruit and S-R relationship Surplus production
iii ) Deterministic analysis
Yield per recruit combined with S-R function
Yield per recruit (assuming recruitment is independent of SSB) All evaluations should include sensitivity analysis of (see WKFRAME I for further discussion):
Choice of S-R functional form Growth and Maturation Density dependence Stability of fishery selection
Choice of period for S-R, growth maturation & fishery parameters
Where there is uncertainty in any aspect it is preferable to include this uncertainty in the analysis carried out, giving appropriate weight to differing possibilities with the objective of defining the ‘best’ target (or range of target exploitation). If natural mor- tality variability is included in the assessment then this should also be included in the evaluation, i.e. it is not considered appropriate to add variability in natural mortality as an additional source of variability to assessment with fixed Ms.
